On 27/08/17 20:53, Clemens Gruber wrote:
Hi,
On Mon, Aug 14, 2017 at 09:42:24PM +0100, Ian Jamison wrote:
Hi,
On 14/08/17 19:38, Clemens Gruber wrote:
(snip)
About the underlying problem (b) why it only occurs on SMP systems:
I think Ian's theory is correct:
DMA is started, then the PIO is done until the xmit buffer is empty and
immediately after that, DMA is stopped.
On SMP systems, where the DMA TX thread can run on another core, it is
already too late.
Regarding problem (a) why it only hurts RS-485: One possibility could be
the timing difference / additional delay due to for example toggling the
transmit-enable GPIO via mctrl_gpio_set.
Meaning that with RS-232 on SMP systems DMA is also stopped just early
enough to not bork the circular xmit buffer.
If this is true then the imx driver did not really use TX DMA in
practice before.
Thoughts?
I'll try to trace this next week to verify these hypotheses.
This sounds plausible to me. I guess you could try adding a GPIO wiggle
in non-RS485 mode (or even a usleep) to see if it makes the problem
more noticeable.
Just a heads-up from my side: I was so far unsuccessful in reproducing
the error on RS-232 interfaces, even when adding large mdelays of 50ms
at all places where we would call imx_port_rts_(in)active in the RS-485
case.
I had broken my build for a while there, but am now testing 4.13-rc5
with RS485 mode and DMA enabled and it seems OK since my v1 patch
is included now. I also tried removing my change to the while loop and
adding a return before it (if dma_is_txing), as Uwe suggested, and that
also seems to work fine. My tests are not very exhaustive though. I
can post that as a patch if you like, or I can test your v2 if you prefer.
Yes, please post it. I also tested that variant, Uwe suggested,
successfully.
It's probably still a good idea to continue the search for the root
cause and explanation why this does not occur with RS-232.
Next step is to trace the functions when doing RS-232 vs. RS-485. Time
for ftrace..
*To be continued* ;-)
Thanks for the continued work on this Clemens. I'll post my patch which
has the commit subject
"serial: imx: Avoid post-PIO cleanup if TX DMA is started"
All the best,
Ian
--
To unsubscribe from this list: send the line "unsubscribe linux-serial" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
